[Hackathon] Frontend runner uses a fixed desktop viewport - cannot test mobile/responsive layouts

[lxcario]

Summary

The frontend runner appears to execute at a fixed desktop viewport width, and there is no way (via a CLI flag or a plan step) to run a test at a mobile viewport. Responsive / mobile-only UI — e.g. a md:hidden bottom navigation bar — therefore can't be verified.

Environment

• CLI: testsprite (latest)
• Test type: frontend (--plan-from)
• Target: live Next.js app using Tailwind responsive breakpoints

What happened

I wrote a plan whose first step was "resize the browser to 390px width" to test a mobile bottom-nav. Across two runs:

• The runner ignored the resize step and instead appended ?mobile=1 to the URL (a no-op for the app).
• The desktop sidebar stayed visible; the md:hidden bottom nav never rendered.
• The mobile layout was never exercised, so the run couldn't assert on it.

Expected

A way to set the viewport / device for a frontend run — for example a --viewport <WxH> or --device <name> flag on test create / test run, or a recognized "set viewport" plan step that the runner honors.

Actual

No viewport / mobile / device / width / height option exists on any command (checked testsprite --help and test create/run --help, and the CLI source). The runner defaults to a desktop viewport with no override, so mobile-responsive assertions are untestable.

Impact

Any app with responsive or mobile-specific UI can't be covered. We had to delete an otherwise-valid mobile bottom-nav test and document it as a runner limitation rather than hide a red result.

Suggested fix

Add a viewport / device override (a flag, or a plan step the runner honors) so responsive layouts can be tested at defined widths.

---

Filed as part of TestSprite S3 Hackathon dogfooding (project: Nora — https://github.com/lxcario/Nora). Happy to provide the exact plan JSON and run IDs.

[lxcario]

Maintainers: could you please add the hackathon and needs-triage labels to this issue? It was filed via gh --body-file, which bypasses the issue-template auto-labeling, and external contributors can't self-apply labels. Thanks!

[zeshi-du]

Confirmed: there is currently no viewport/device override anywhere in the pipeline, and the agent improvising around an un-honorable resize step (the ?mobile=1 URL trick) is its own bug. Both are now tracked internally — the override needs plumbing from the run-trigger API through the browser engine and replay, so it's a cross-layer feature rather than a quick CLI flag. Leaving this open as the public tracking issue.